13
Package-Based Description Logics
Jie Bao
1
, George Voutsadakis
2,3
,GioraSlutzki
2
, and Vasant Honavar
2
1
Department of Computer Science, Rensselaer Polytechnic Institute, Troy, USA
baojie@cs.rpi.edu
2
Department of Computer Science, Iowa State University, Ames, USA
{slutzki,honavar}@cs.iastate.edu
3
Department of Computer Science, Lake Superior State University, USA
gvoutsad@lssu.edu
Summary. We present the syntax and semantics of a family of modular ontology languages,
Package-based Description Logics (P-DL), to support context- specific reuse of knowledge
from multiple ontology modules. In particular, we describe a P-DL SHOIQP that allows
the importing of concept, role and nominal names between multiple ontology modules (each
of which can be viewed as a SHOIQ ontology). SHOIQP supports contextualized inter-
pretation, i.e., interpretation from the point of view of a specific package. We establish the
necessary and sufficient conditions on domain relations (i.e., the relations between individuals
in different local domains) that need to hold in order to preserve the unsatisfiability of concept
formulae, monotonicity of inference, transitive reuse of knowledge across modules.
13.1 Introduction
The success of the world wide web can be partially attributed to the network effect:
The absence of central control on the content and the organization of the web al-
lows thousands of independent actors to contribute resources (web pages) that are
interlinked to form the web. Ongoing efforts to extend the current web into a se-
mantic web are aimed at enriching the web with machine interpretable content and
interoperable resources and services [7]. Realizing the full potential of the semantic
web requires the large-scale adoption and use of ontology-based approaches to shar-
ing of information and resources. Constructing large ontologies typically requires
collaboration among multiple individuals or groups with expertise in specific areas,
with each participant contributing only a part of the ontology. Therefore, instead of
a single, centralized ontology, in most application domains it is natural to have mul-
tiple distributed ontologies covering parts of the domain. Such ontologies represent
the local knowledge of the ontology designers, i.e., knowledge that is applicable in
a context. Because no single ontology can meet the needs of all users under every
conceivable scenario, there is an urgent need for theoretically sound, yet practical,
H. Stuckenschmidt et al. (Eds.): Modular Ontologies, LNCS 5445, pp. 349–371, 2009.
c© Springer-Verlag Berlin Heidelberg 2009

350 J. Bao et al.
approaches that allow knowledge from multiple autonomously developed ontologies
to be adapted and reused in user, context, or application-specific scenarios.
Ontologies on the semantic web need to satisfy two apparently conflicting objec-
tives [9]:
• Sharing and reuse of knowledge across autonomously developed ontologies. An
ontology may reuse another ontology by direct importing of selected terms in the
other ontology (e.g., by referring to their URLs), or by using mappings between
ontologies.
• The contextuality of knowledge or accommodation of the local points of view.
For example, an assertion of the form “everything has the property that...” is
usually made within an implicit local context which is often omitted from the
statement. In fact, such a statement should be understood as “everything in this
domain has the property that...”. However, when reusing an existing ontology,
the contextual nature of assertions is often neglected, leading to unintended in-
ferences.
OWL adopts an importing mechanism to support integration of ontology modules.
However, the importing mechanism in OWL, implemented by the owl:imports
construct, in its current form, suffers from several serious drawbacks: (a) It directly
introduces both terms and axioms of the imported ontologies into the importing onto-
logy, and thus fails to support contextual reuse; (b) It provides no support for partial
reuse of an ontology module.
Consequently, there have been several efforts aimed at developing formalisms
that allow reuse of knowledge from multiple ontologies via contextualized inter-
pretations in multiple local domains instead of a single shared global interpretation
domain. Contextualized reuse of knowledge requires the interactions between local
interpretations to be controlled. Examples of such modular ontology languages in-
clude: Distributed Description Logics (DDL) [8], E-Connections [16] and Semantic
Importing [20].
An alternative approach to knowledge reuse is based on the notion of conser-
vative extension [12, 13, 14, 15], which allows ontology modules to be interpreted
using standard semantics by requiring that they share the same global interpretation
domain. To avoid undesired effects from combining ontology modules, this approach
requires that such a combination be a conservative extension of component modules.
More precisely, if O is the union of a set of ontology modules {O
1
, ..., O
n
},then
we say O is a conservative extension of O
i
if O |= α ⇔ O
i
|= α,foranyα in the
language of O
i
. This guarantees that combining knowledge from several ontology
modules does not alter the consequences of knowledge contained in any component
module. Thus, a combination of ontology modules cannot induce a new concept in-
clusion relation between concepts expressible in any of the component modules.
Current approaches to knowledge reuse have several limitations. To preserve con-
textuality, existing modular ontology languages offer only limited ways to connect
ontology modules and, hence, limited ability to reuse knowledge across modules.
For instance, DDL does not allow concept construction using foreign roles or con-
cepts. E-Connections, on the other hand, does not allow concept subsumptions across